THE JOURNAL OF INDUSTRIAL ECONOMICS

Volume XLVII
June 1999

0022-1821
No. 2

MARKET STRUCTURE, R&D AND ADVERTISING IN THE

PHARMACEUTICAL INDUSTRY*
CATHERINE MATRAVEsf

Recent developments in the literature on market structure have

allowed the generation of a few key testable predictions from the
theory of strategic behaviour. The seminal model considers one simple
but general relationship, that between market structure and market
size, focusing on the competitive roles of endogenous sunk costs in the
form of advertising and/or research and development (R&D).
Evidence presented in this case study, building on earlier econometric
work, shows that such endogenous sunk costs do play a crucial role in
the formation of market structure in the global pharmaceutical
industry.

I.

INTRODUCTION

of game theory has allowed large advances to be made

within industrial economics, but it has become clear that game-theoretic
models often yield extremely sensitive predictions. Furthermore,
unobservables such as the nature of inter-firm competition, information,
etc., make the models hard to operationalise as it is very difficult to
identify any market characteristic that can act as an adequate proxy.
Precisely because there exist many real world strategies, there is no reason
to expect a single oligopoly theory that would deliver universal
predictions. Yet for the theory to be useful, it must be helpful either in
analysing particular industries or in identifying behavioural cross-industry
regularities (Shapiro [1989]). Recent empirical work has tended to focus
on the former, showing that firm conduct and performance in a particular
market can sometimes be explained using some chosen game-theoretic
model. Representative examples include Bresnahan [1981], Slade [1987],
Hendricks, Porter and Boudreau [1987]. However, these ultra-micro
studies remain unable to explain widespread empirical regularities.
THE DEVELOPMENT

* An earlier version formed part of my PhD dissertation at the University of East Anglia,
Norwich, UK. I would like to thank Bruce Lyons and Steve Davies for their valuable
suggestions. The paper also benefitted considerably from the comments of Lars-Hendrik
Roller, Bernard Sinclair-Desgagne, Mike Waterson, and two anonymous referees. All
remaining errors are mine.
t Author's affiliation: WZB, 10785 Berlin, Germany; and Department of Management,
Texas A&M University, College Station, TX 77843-4221, USA.
email: cmatraves@cgsb.tamu.edu,
O Blackwdl Publiihen Ltd. 1999. 108 Cowlcy Road. Oxford 0 X 4 UF. UK. and 3S0 Main Street. Maiden. MA 02148. USA.

169

170

CATHERINE MATRAVES

A seminal contribution that breaks with this new tradition is Sutton

[1991]. His 'lower bound' approach eschews detailed and sensitive gametheoretic results, but generates a few key robust and testable predictions
from the theory of strategic behaviour. Sutton emphasises the strategic
choices of sunk costs in a simple and highly general framework, focusing
on one simple relationship, the relationship between market structure and
market size. He uses cross-section econometric analysis and industry case
studies to test his predictions, where the latter provide rather more
qualitative evidence on the underlying competitive mechanisms. Homogeneous and advertising intensive (vertically differentiated) industries are
analysed. The general empirical results obtained by Sutton usefully
complement and extend the ultra-micro studies: 'the experiences of
individual industries can be mapped into special cases of a general
theoretical model whose robust results drive the cross-industry
regularities' (Sutton [1991, p. 321]).
This paper has two main objectives. First, this industry study
complements and reinforces earlier cross-section econometric work
(Sutton [1991]; Lyons and Matraves [1996]; Robinson and Chiang
[1996]; Lyons, Matraves and Moflfatt [1998]). Specifically, by concentrating on the experience of an individual industry, we add to the case
evidence accumulated for major national markets in the food and drink
industry as detailed in Sutton [1991]. An industry study allows a better
analysis of the industry dynamics, illustrating how exogenous structural
changes feed into the competitive process. Using evidence on observed
market share data, we can infer how the underlying competitive
mechanisms operate in the pharmaceutical industry. Secondly, in the
new theory of industrial structure, the distinction between horizontal
and vertical differentiation is crucial. In industries in which vertical
differentiation matters, advertising or R&D expenditure is effective in
increasing consumers' willingness to pay. Interestingly, the pharmaceutical industry is both advertising and R&D intensive, allowing the
analysis of the relative infiuences of competition in two types of
endogenous simk costs.
Section II sets out the theoretical framework, based on Sutton [1991,
1996, 1998]. Production economies of scale, R&D and advertising each
play a central role in the story. We characterise the pharmaceutical
industry according to the industry typology developed. In Section III, the
recent structural changes that have taken place are discussed, and then fed
into the theoretical framework in order to derive predictions. In Section
IV, it is demonstrated that the industry level evidence is consistent with
the predictions derived, using national concentration data for the period
1987-1993 and global market share data for 1983-1995 (the latest years
for which such data could be derived). The final section offers some concluding remarks.
O Blackwell Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

II.

171

THEORETICAL FRAMEWORK: THE SUTTON APPROACH

Following the classification of Sutton [1991], manufacturing industries

can be split into two types, which Schmalensee [1992] labelled Type 1 and
Type 2. Type 1 industries are characterised by homogeneous and
horizontally differentiated products. Type 2 industries are characterised by
vertically differentiated products.
In Type 1 industries, production economies of scale, market size, price
competition and entry conditions combine to determine market structure.
For a class of two-stage games, it is shown that there is an inverse
structure-size relationship.' The basic intuition is that as market size
increases, profits increase, and given free entry, firms enter until the profits
of the last entrant just cover the exogenous overhead costs paid on entry.
The more sensitive price competition is to entry, the bigger is the gap
between pre-entry and post-entry profits which reduces the extent of entry.
Overall, the net effect of an increase in market size must be a rise in firm
numbers and reduced concentration. Allowing horizontal differentiation
does not substantially alter the argument, except that a range of outcomes
becomes possible. Small niche firms may be able to survive as well as
multi-product firms who fill many of the gaps and capture a larger market
share than in the single product case.
In Type 2 industries, firms not only compete in price and horizontal
product differentiation, but also in their advertising and/or R&D
expenditure. In general, such expenditure is a choice variable, and it is this
choice of (perceived) quality that highlights the fundamental difference
between Type 1 and Type 2 industries.^ Firms can react to increases in
' As a simple example of a Type 1 industry, assume the total value of market demand is
fixed at S (i.e. there is unit elasticity of demand). Straightforward calculations show that if all
Af firms in the market have the same constant marginal costs, and competition is Coumot,
gross profit will be 7t = S/N^. If each firm must incur a fixed cost, a, upon entry, the free entry
zero-profit condition (n = a) gives the long run equilibrium number of finns: N = [S/a]'^,
^ As a simple example of a Type 2 industry, again assume market demand is fixed at S,
and allow consumer utility to depend not just on quantity consumed, but on the product of
quantity and u, where u is an index of (perceived) product quality. Sutton [1991, Ch. 3]
proposes the following relationship between endogenous sunk costs, E, and u:
E(u) = [a/y][u' 1] where y > 1; a higher y corresponds to more rapidly diminishing returns
to investment; a reflects the unit cost of investing. Given strictly positive E, and noting that in
equilibrium firms will invest until
du

dE
du

the symmetric subgame perfect Cournot Nash equilibrium with free entry (n = S/N^ = E + a)
generates

In the limit, as S ^ co, [N + N ]-* (2 + (y/2)] which is a finite constant. Furthermore,

sgn [dN/dS] = sgn{a - (a/y)), which confirms the possibility that concentration may actually

rise with market size.

O Blackwell Publishers Ltd. 1999.

172

CATHERINE MATRAVES

market size by investing in enhancements to (perceived) product quality

and so typically increasing consumers' willingness to pay. Intuitively, as
market size increases, the incentive to gain market share through the
escalation of advertising and/or R&D expenditure also increases, and the
consequent rise in overhead costs has a countervailing effect on market
structure by increasing the degree of economies of scale. The basic notion
therefore is that although there appears to be room for more firms as
market size increases, the 'escalation mechanism' raises fixed costs per firm,
possibly even to such an extent that the negative structure-size relation
breaks down. Certainly, market structure will be less fragmented than in
Type 1 industries.
Within the set of Type 2 industries, Sutton [1996, 1998] further analyses
the relationship between the nature of technology and market structure
to explain why some R&D intensive industries remain unconcentrated. He
argues that it is not always the case that in an innovative Type 2 industry,
firms move forward along one single technological path making
successively better products. Instead, various technologies can be adopted,
where a technology is the chosen route for the R&D expenditure. Each
technology may lead to several marketable products, where these products
may be close or distant substitutes for products associated with another
technology. A firm may choose either to: (i) focus on a few technologies,
undertaking a lot of R&D per technology; or (ii) spread its R&D
expenditure across a wide range of technologies, (i) is associated with
the standard 'escalation mechanism' as previously described. Such
industries are termed 'Type 2 high-a'. (ii) is associated with the
'proliferation' of various distinct technologies, and such industries are
termed 'Type 2 low-a'.^
The value of a will be higher the cheaper it is to improve product
performance as this encourages escalation rather than entry. The value of
a will also be higher if there is only one technology, or the closer the
'As a simple example (Sutton [1996]), assume a standard linear demand model, extended
to allow goods to differ in quality. Thus,

where M = money spent on outside goods; Uj > 1 = the quality of good k; and 6 the degree
of substitution between goods where if S = 1, the goods are perfect substitutes. Assume each
firm adopts a single technology and has a single product. Following Sutton [1991] as in
footnote 2, but using a slightly more restrictive functional form for E{u) = u' where y is the
cost (or alternatively, the effectiveness) of R&D in raising product quality (performance), the
subgame perfect zero profit equilibrium yields

For this example, the limiting level of concentration is determined by the values of y and 6.
Thus, for a given Q, a lower y implies higher concentration.
O Bbclcwcll Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

173

substitutability of rival technologies (and associated product groups), as

a deviant firm that escalates R&D along its chosen technological path can
gain market share from firms that spend less in other technologies.'*
The crucial theoretical notion therefore is the share of industry sales that
can be achieved by a deviant firm that outspends all its rivals along a
particular technological path and offers a 'higher-quality' product. In a
Type 2 high-a industry, such a firm can capture some given market share,
independently of how many other firms are in the market. Thus, Type 2
high-a industries will be associated with high R&D and high
concentration. On the other hand, in Type 2 low-a industries, if products
are distant substitutes (i.e., a large degree of horizontal differentiation),
then a firm that outspends its rivals along a particular technological path
can only capture sales for a small product group. Such a firm will not be
able to capture market share from rival firms on different paths, and
therefore, the incentive to introduce a new variety, i.e., to proliferate, is
stronger.^ This implies lower concentration, even if R&D expenditure is
high. In Type 2 low-a industries, the escalation of R&D leading to a
shakeout of technologies over time will not be observed, but rather a
proliferation of technologies, where new ones enter and exist alongside the
old.
Characterising the pharmaceutical industry within this framework^

Whether concentration is high in an R&D intensive industry such as the

pharmaceutical industry essentially depends on the substitutability of
products associated with different technologies. If substitutability is low,
then in spite of the effectiveness of R&D, concentration may also be low.
In the pharmaceutical industry, there exist many therapeutic classes (see
Appendix), and a wide range of technologies. Indeed, Temin [1979, p. 444]
argues that this industry was transformed after the Second World War
by a new research methodology, but there is no one central product: 'the
revolution in drug research allowed many different drugs to be discovered
and promoted.' A firm's product range often tends to be horizontally
differentiated, making various formulations of the successful drug rather

"The value of a tells us the extent to which increases in R&D expenditure would increase
a deviant firm's market share. For the example in Footnote 3, a is a function of y and 6,
neither of which can be directly measured. Empirically, the value of a can be linked to the
R&D to sales ratio and the number of different technologies (which must be proxied).
' Inasmuch as there exist any economies of scope in R&D, then for a given market size, this
will shift up the lower bound to concentration.
* Advertising/R&D to sales ratios of 1% provide suitable empirical criteria for classifying
an industry as Type 1 or Type 2. Importantly, these data are used purely to give a 'broadbrush' distinction between those industries identified by the exogenous sunk cost model (e.g.
bread, salt) versus the endogenous sunk cost model (e.g., cars, soft drinks).
O Blackwdl Publishen Ltd. 1999.

174

CATHERINE MATRAVES

than successive improvements. Each new drug must attain a certain

quality standard that is set by the regulatory authority, and differences
across firms (see Section IV) tend to appear more in the number of
therapeutic classes covered. Thus, we characterise the pharmaceutical
industry as a Type 2 low-a industry. R&D expenditure is among the
highest in any manufacturing industry (indicating high R&D effectiveness)
and yet concentration is low.^
An interesting criticism of the industry characterisation (Type 2 high-a
or Type 2 low-a) is related to the issue of market definition. It can be
argued that once the 'right' market has been defined, high R&D and high
concentration would be observed. However, in regard to the pharmaceutical industry, this argument is fiawed.* For example, although
manufacturing capacity is easily transferable once the technology is
known, the degree of substitution (on the demand side) between and within
therapeutic classes is not at all easy to calculate (Caves, Whinston and
Hurwitz [1991]). First, one drug might be the typical choice for a
particular illness, but not for a certain subset of patients in whom side
eflfects occur. Secondly, there may be several drugs available for a given
illness, where the actual choice depends on trial and error to see which one
is best for the patient. Thirdly, one drug used for several illnesses may have
different substitutes in each use. Finally, drug substitutability does not
remain constant over time as new entrants appear, relative prices change,
and prescribers' preferences change according to accumulated empirical
data on safety and efficacy (Caves, Whinston and Hurwitz [1991]). These
difficulties in market definition are due to the splintering of the incentives
in R&D (escalation or proliferation), which is in itself endogenous (Sutton
[1996]).

'Mean concentration for industries characterised by both intensive advertising and R&D
expenditure was 44% in the US and 40% in the EU market in 1987 (Lyons, Matraves and
Moffatt [1998]). Table VI shows that concentration is much lower in the pharmaceutical
industry: 26% for the US, 19% for the EU.
* Note that average concentration is substantially higher at the therapeutic class level. Cool,
Roller and Leieux [1998] estimated mean US concentration to be 46.3% in 1982, and also
showed that this was relatively stable between 1963-1982. However, this stability may mask
significant turbulence (changes in survivor market shares and new drug introductions).
Unfortunately, global market size and market share data were unobtainable by therapeutic
class. One example of turbulence, however, is in anti stomach ulcer drugs, where the exbestseller, Zantac, had a 42% global market share until its patent expired. The pioneer,
Tagamet, was introduced in 1977. Zantac was marketed globally in 1983 and became the
market leader by 1987; Zantac had fewer side effects than Tagamet, although Tagamet was
also considered very safe. Two other H2-blockers were introduced in the late 1980s and did
not overtake Zantac. In 1989, a new type of anti-ulcer drug, Losec (a proton pump inhibitor
rather than a H2 antagonist) was launched; by 1994, Losec's sales were $2.2 billion compared
to Zantac's $3.7 billion. It appears that the 'exclusivity period' is decreasing (Grabowski
and Vemon [1990]; also see PhRMA [1996], for more examples).
e Blackwdl Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

III.

175

STRUCTURAL CHANGE: ANY PREDICTIONS?

This section considers recent structural changes in the pharmaceutical

industry which have to be considered in some detail in order to derive
predictions in the light of the theoretical framework set out in Section II.
First, the impact of globalisation is analysed, which is defined to include
both structural forces and firms' actions. In particular, the Single
European Market (SEM) legislation and other regulatory changes are
taken into account. The predicted impact on concentration of this increase
in global market size, and tougher price competition, are derived.
Secondly, the potential effects of the changing discovery process (towards
rational drug design) and of the cost of innovation on industry
concentration are assessed. Finally, the relative importance of the overthe-counter (OTC) sector has been increasing in value terms in recent
years. This may make advertising more effective, and consequently, have
an impact on inter-firm competition, and hence, concentration.
Globalisation, Regulation and Price Competition

Over the past decade, leading firms have been competing more and more
at a 'global' level. Moving out of their home markets, particularly
European firms entering the US in the 1980s, these firms produce and
market their products worldwide, though access to the Japanese market
remains relatively more difficult due to extensive non-tariff barriers
(Thomas [1996]).' Most of the world's Top 20 firms, with approximately
50% of global pharmaceutical sales and 85% of R&D expenditure, operate
in the key markets of North America, Europe and Japan (Grabowski and
Vemon [1994a]). The main driving forces of this globalisation trend are
outlined below.
First, competition is tougher due both to the emergence of new
technologies at the global level, and higher R&D costs. Even if the pricing
choice remains region specific, a firm's sunk cost decision may be taken
with the global market size in mind, depending on the transferability of
the sunk cost across borders (Davies and Rondi [1996]). Unlike advertising
expenditure, which is typically dependent on national culture, media and
language, the results for R&D are relatively easily transferable across
national borders. Once an innovation has been made, it can be exploited
'Pharmaceutical exports for all OECD countries increased from 23.5% of total production
in i982 to 28.5% in 1992. Sharp and Patel [1996] indicate the degree of self-sufficiency in the
key markets for 1991: Japan was the highest at 82% (German firms took 5%), the US with
70%, Germany with 52%, the UK with 35%. Three European countries have a significant
market share in the US market: UK firms with 15%, Swiss firms with 8%, and German firms
with 4.6%; US firms had a market share of 19% in Germany, 26% in the UK, and 21% in
France.
O Blackwell Publishers Ud. 1999.

176

CATHERINE MATRAVES

anywhere in the world, although some modification or additional testing

of the new product may be necessary to conform to local regulations.
Additionally, the need to market a product effectively is reinforced by the
increase in the demand for OTC products. The leading firms have a
presence in all key markets, either through their own marketing and
distribution networks, or via a marketing joint venture. Multinationality is
thus one indicator, rather than a more conventional trade measure, of
the extent of integration."*
Secondly, although the pharmaceutical industry remains very highly
regulated in the key markets following the thalidomide disaster in the
1960s, there have been important movements towards increased
international market harmonisation. In the EU, the SEM legislation
removed technical barriers to trade, and allowed firms to access the
regional markets more easily. From 1st January 1995, the European
Medicines Evaluation Agency (EMEA) has recommended drugs for EUwide circulation, reported any adverse reactions, and coordinated
inspections." Firms that supply a local market only negotiate with
national agencies. Furthermore, the International Conference on Harmonisation (ICH) was set up in 1990 to bring together regulatory authorities
and drug developers from the US, the EU and Japan to internationally
harmonise regulatory procedures. In 1996, for example, an important
agreement was reached by the ICH on the appropriate standards for the
good conduct of clinical trials. This was implemented in 1997, and means
that clinical trials undertaken in the UK, say, will no longer have to be
replicated in the US (Internet ICH publication) if the appropriate
guidelines are followed. Various other guidelines relating to quality, safety
and efficacy such as the nixmber of patients in clinical trials, the length of
exposure to the drug, etc., have also been agreed on, and in some cases,
implemented.
Thirdly, the common mounting governmental concern over increasing
healthcare costs is having a negative impact on pharmaceutical prices, via

'"indeed, our analysis of leading European pharmaceutical firms in 1993 strongly indicates
the importance of foreign markets. One example is Glaxo: sales by origin were 21% in the
UK, 21% in the rest of the EU, 44% in the US and 14% in the rest of the world (details of
other firms are available on request from the author). Ballance et al. [Ch. 3, 1992] also argue
foreign direct investment is more important in this industry than trade.
" In the US, the Food and Drug Administration (FDA) approves new drugs, where both
safety and efficacy must be shown; in Japan, safety only has to be proven, not efficacy. EU
requirements are generally not as tough as in the US (Dranove [1991]), and US approval costs
tend to be higher (Senker, Joly and Reinhard [1996]). Of the 154 drugs approved by the
FDA between 1990 and 1995, 103 (67%) were first approved outside the US (PhRMA [1996]).
In 1992, the Drug User Fee was passed which allowed the FDA to hire more reviewers to
speed up the approval process. In 1987, the mean review time was 32.4 months with 21 drugs
approved. This had decreased to 17.8 months by 1997 with 53 drugs approved. The statutory
standard is six months.
O Blackwell Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

177

TABLE 1
MARKET SIZE AT CURRENT PRICES (SBILLION)

Market Size

GER

FR

UK

EU

US

Japan

Global

1987
1988
1989
1990
1991
1992
1993

11.8
12.9
13.3
17.1
17.9
20.3
19.5

10.2
11.6
11.8
14.8
15.0
17.3
17.1

8.2
10.2
10.4
12.1
13.3
15.1
14.9

47.7
55.4
59.4
74.9
80.2
90.2
88.8

39.3
44.0
49.1
53.7
60.8
61.1
70.8

30.2
36.2
36.4
35.7
39.8
43.7
51.1

135.5
156.6
168.0
190.2
208.0
229.9
239.5

Source: These 'value of production' data are taken from the OECD 'Indtistrial Structure Statistics' and
account for ethical Pharmaceuticals only. The dollar conversion was made using the nominal average
exchange rate (OECD Outlook, June 1994).

reducing reimbursement rates and increasing the 'limited lists'. This

increases the incentive to market the product globally. In the US, there is
very little direct price intervention. Price competition has recently however
been influenced by the rapid expansion of health care maintenance
organisations (HMOs) which has concentrated drug purchasing, and had
the effect of reducing the number of drugs prescribed to only one or two
per indication. Virtually all HMOs use limited lists, or so-called
formularies, and by 1995, such organisations accounted for 75% of US
drug purchases. On the other hand, in the EU and Japan, where the
government is the main purchaser, there is substantial price intervention
of one form or another.
The continuing globalisation process increases effective market size as
leading firms have access to new geographical markets. Table I shows the
increase in actual market size due to factors such as rapidly ageing
populations in most advanced industrialised nations, and the substitution
of drugs for more expensive forms of healthcare such as hospitalisation/
surgery. EU pharmaceutical production, for example, increased at an
average annual growth rate of 9.7% (5% in real terms) between 1987 and
1993. Between 1984 and 1993, the increase in the value of production
(when measured in constant prices) was 46% for the EU, compared with
30% for the US, and 21% for Japan (Panorama [1995]). As market size
increases, this raises the incentive to escalate advertising and/or R&D
expenditure (Sutton [1991]).'^ This yields Prediction 1.
Prediction 1: An increase in market size is associated with an increase in
the level of advertising and/or R&D expenditure by each surviving firm.
Consider also the toughness of price competition. Sutton [1991] shows
'^ In Type 2 industries in which the escalation mechanism comes into play, this will increase
concentration if the escalation effect outweighs the direct market size effect.
O Blackwdl Publishers Ltd. 1999.

178

CATHERINE MATRAVES

that in Type 1 industries, any exogenous influence that has a negative

impact on prices or gross profit margins increases the toughness of price
competition. If price competition becomes tougher due to governmental
policies or an increase in imports/foreign direct investment, then the result
is increased concentration. In Type 2 industries however this relationship
can be ambiguous. In particular, Symeonidis [1997a, 1997b] shows that the
effect of the toughness of price competition on the level of concentration
is theoretically indeterminate. Essentially, whether concentration increases
or decreases depends on the relationship between endogenous sunk costs
and price competition. He shows that as price competition becomes
tougher, if advertising/R&D increases or is constant, then concentration
must increase. A case exists however where if advertising/R&D decreases
with the intensity of price competition, then this may or may not offset the
fall in gross profit caused by lower prices, so concentration may fall or rise.
Symeonidis [1997a] looks for empirical regularities, using concentration
data at the 4-digit industry classification level post-implementation of the
UK Restrictive Trade Practises Act in 1956 (previously allowable
restrictive agreements between manufacturing firms were prohibited). He
funds that as the toughness of price competition increased, concentration
increased irrespective of whether the industry was characterised by
exogenous or endogenous sunk costs.'^ This yields Prediction 2.
Prediction 2: An increase in the toughness of price competition implies
an increase in concentration.
The R&D Process and the Cost of Innovation

There has been a radical shift in the methodology by which new drugs
are discovered. The initial question arises whether this shift has affected
the underlying competitive mechanisms in the pharmaceutical industry.
Furthermore, the average real cost of introducing a new drug, which we
label the 'cost of innovation', has increased. The factors leading to this
increase are discussed, and the resulting predicted impact on the level of
concentration is derived.
Historically, the technologies by which most drugs were discovered can
be traced to random screening of thousands of compounds for efficacy
against a given disease, accidental discoveries, or incremental improvements to existing drugs (Schwartzman [1976]). Over the past two decades,
'random drug design' has been replaced by new technologies applying the
more focused 'rational drug design'. Due to major advances in basic
"Even if advertising/R&D expenditure decreases, in order for concentration not to rise,
it must be a significant decrease to offset the reduction in profits arising from tougher price
competition. The most likely outcome is an increase in concentration.
O Blackwdl Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

179

biomedical knowledge, scientists are now better able to understand

biological interactions in the body, and hence design specific drugs to
affect these interactions. Since the mid 1980s, drugs have been increasingly
manufactured from biotechnological processes rather than from chemical
ones. Although the results from biotechnology can be applied within the
pharmaceutical industry, the actual scientific advances were achieved
outside the pharmaceutical industry and thus exogenous.
As explained in Section II, the value of a depends on the effectiveness
of R&D and the substitutability of products associated with different
technologies. If a new technology is introduced which increases consumers'
willingness to pay, then in a Type 2 high-a industry, the escalation
mechanism comes into play. Shakeout occurs as firms race for market
share, and the lower bound to concentration will be higher. New
technologies displace old ones.
In a Type 2 low-a industry such as Pharmaceuticals, the value of a may
not change because even if biotechnology does become the dominant
discovery process, there are many research trajectories (and associated
therapeutic classes) over the entirety of the pharmaceutical industry.'''
This implies that the value of a is not likely to significantly increase, as
escalation along one technological trajectory will not cause firms to exit on
other trajectories. In other words, the shift from random to rational drug
design in the pharmaceutical industry may act as a natural experiment. In
a Type 2 high-a industry, we would predict shakeout and higher
concentration in the new equilibrium.'^ In a Type 2 low-a industry, on the
other hand, small and mid-sized firms may still be able to survive by
becoming more specialised. Suppose leading pharmaceutical firms are
active in many therapeutic classes. Not only do these firms incur set up
costs to enter each segment (if plant costs are therapeutic class specific)
but they also incur the costs necessary to develop a new drug (the cost of
innovation). Given the likelihood of capital constraints for mid-sized
firms, it may be optimal to devote all R&D efforts to one technology only
to capture a large global market share for that particular product line.
Although the new drug discovery process is more efficient, the 'cost of
innovation' has increased. The average cost of developing a new drug, or
the cost of innovation, was estimated to be $359 million in 1992 (taking

'For example, Penan [1996] identifies 15 distinct research programs studying Alzheimer's
disease alone. These work through two main R&D technologies: neurotransmitter
augmentation and cognitive enhancing agents to slow down the neurodegenerative process.
Henderson and Cockburn [1996a] assert that leading firms typically invest in 10-15 different
research programs, where each program is targeted towards a particular disease area.
" Sutton [1996] uses the colour film market as an example of a Type 2 high-a industry
where during the shift from black and white to colour film, R&D expenditure escalated, and
merger/exit took place. Today, the global market is dominated by just five firms.
<0 Blackwell Publishers Ltd. 1999.

180

CATHERINE MATRAVES

into account compounds that failed),'* compared with $231 million in

1987, only five years earlier, and $54 million in 1976 (Hansen [1979]; Di
Masi et al. [1991]; Pisano and Wheelwright [1995]). The factors underlying
the increase in the cost of innovation are varied. First, if we think of the
underlying technological opportunity, then it may be as if the 'easy' drugs
have already been developed, i.e., moving away from 'easier' therapeutic
areas to areas such as oncology or gerontology (Henderson and Cockburn
[1996b]). However, their evidence for 10 leading pharmaceutical firms
suggests this is probably not the case. Secondly, even if the discovery
process is more efficient, if new drugs are more difficult to find due to the
complexity of the compounds, it follows that the costs of doing so are
higher. Thirdly, pharmaceutical research is, nowadays, more focused on
chronic and degenerative illnesses rather than acute illnesses, mainly due
to rapidly ageing populations. Such illnesses tend to require longer and
more costly clinical trials (Di Masi et ai [1991]). Finally, regulatory
requirements are far tougher regarding the safety and efficacy of a new
drug.''' Lengthening development times in clinical trials increase the cost
of capital required to undertake R&D. As Table II shows, the introduction
of new products (NCEs) is slowing. In 1996, only 37 NCEs were launched
(Financial Times, 24/4/97), which was the lowest total for several
decades.'^
The increase in the cost of innovation sets a lower limit on the size of a
firm, and this size effect is further reinforced by the following two stylised
facts (Grabowski and Vernon [1990, 1994b]): (i) the return to new drugs
is highly skewed, a few 'blockbusters' dominate the product ranges of the
major firms;" and (ii) only the top 30 drugs worldwide cover average
R&D costs. Interestingly, smaller firms are disproportionately affected by
both of these factors. In the long run, firms must cover total fixed costs or
exit/merge. Therefore, even if there is variance in the cost of innovation
across therapeutic classes, smaller firms will be at a disadvantage
"There is significant variance in development costs across therapeutic classes (Di Masi et
al. [1991]; Henderson and Cockburn [1996b]), but there is little evidence of the skewness that
is observed on returns to new drugs (Grabowski and Vemon [1994b]).
"Average drug development time in the US was 8.1 years in the 1960s, 11.6 in the 1970s,
14.2 in the 1980s, and reaching 14.8 years in the 1990s (PhRMA [1996]). It is estimated that
only 1 compound from an initial 5000 will be successfully developed, where clinical trials
account for approximately 35% of total R&D costs (PhRMA [1997])the average number of
trials per new drug application rose from 30 (1977-1980) to 60 (1989-1992).
"since 1960, the majority of NCEs have been discovered in the EU. Japan has the lowest
percentage of NCEs, which may reflect the extremely protected domestic market, although
Japan has overtaken the US in the last decade. However, if the number of breakthrough
NCEs is considered, then between 1970-1983, US firms developed 42%, whereas Japanese
firms only developed 4% (Ballance era/. [1992]; Thomas [1996]).
" i n 1993, Zantac accounted for 44% of Glaxo's sales, Tenormin accounted for 47% of
Zeneca's, Augmentin (22%) and Tagamet (19%) accounted for 41% of SKB's, and Cipro
(23%) and Adalat (20%) accounted for 43% of Bayer's (derived from company accounts).
O Biackwell Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

181

TABLE II
NUMBER OF N C E S INTRODUCED BY COUNTRY OF ORIGIN
Country

1961-70

1971-80

1981-90

W. Europe

509

US

201

Japan
TOTAL
World Total

80
790
844

375
152
75
602
665

243
117
126
486
506

Source: Ballance et al. (1992, p. 86).

Notes: 1961-1980: Italy = 112; Fr = 272;Ger = 191;UK. = 73.1981-1990: Italy = 42;
Ger = 37; Fr = 31; UK = 21 (Earl-Slater, 1993, p. 91). Note that (i) some NCEs
introduced are genuinely original, others may be marginal improvements; (ii) these
data are by country of origin, not firm of origin.

compared with larger firms who may be able to better diversify their
portfolio and absorb the risk of failure. Additionally, in order to recoup
R&D costs, extensive marketing is essential to launch the new drug
effectively, which works against small and mid-sized firms.
Let the fixed and sunk outlay required to achieve quality level, u, be
written as F(u) = 0 + E(u) where 0 = cost of innovation; E(u) = u' as in
our earlier example (footnote 3). An increase in the cost of innovation, <f),
is analogous to an increase in the exogenous overhead costs a firm would
pay upon entry into the market. It has no effect on y (the effectiveness of
R&D). It follows that it has no effect on the asymptotic level of
concentration as market size becomes very large. For any given market
size, however, concentration will increase (Sutton [1991, pp. 79-81]).
Under these conditions.
Prediction 3: As the cost of innovation <j) increases, the minimal level of
concentration for any given market size will increase.
Prescription Drugs versus OTC Products
Prescription drugs that offer a therapeutic advance have a first mover
advantage that tends to be overcome only if later entrants offer a distinct
therapeutic benefit, and not just a lower price (Bond and Lean [1977]).
Even once the patent has expired, pioneer products tend to keep a large
market share due mainly to prescribers' existing familiarity with the
product, and successful marketing (Caves, Whinston and Hurwitz [1991]).
Thus, entry is made difficult on the demand rather than the supply side,
through the stock of goodwill and prescribers' concerns over quality
differences. These effects are now somewhat weaker due to governmental
concern over rising costs forcing doctors to prescribe more cost-effectively
via generic substitution, the increasing importance of HMOs (in the US),
and better information (Grabowski and Vernon [1992]).
e Biackwell Publishers Ltd. 1999.

182

CATHERINE MATRAVES

New products (and processes) are protected by a patent whose typical

life is twenty years. A patent's approximate effective length was, until
recently, ten years only, primarily due to the length of development time in
clinical trials. There have since been changes made to patent law in all
the key markets. In the US, the 1984 Waxman-Hatch Act extended the
patent life for 5 years. Similar patent extension was legislated in Japan in
1986. Within the EU, the 1993 Supplementary Protection Certificate also
extended the patent for up to 5 years. Additionally, in the US, the 1984
Act stipulated that the innovator must provide data to firms who wish to
market the drug as a generic post-patent.^ Thus, although the monopoly
period was increased, the effective time between patent expiration and
generic entry has been reduced to zero. Consequently, the generics' share
of the US prescription drug market by volimie rose from 18.6% in 1984 to
40.4% in 1993. In 1993, generics accounted for approximately 15% of the
EU market, although this varies across member states, and 7% of the
Japanese market (Scrip Yearbook [1995]).
Product life cycles are becoming shorter due to the reduction in the
effective patent life, the increased ability by rival firms to use rational drug
design to produce close substitutes ('me-too' drugs), and generic
competition. In response, leading firms have both offered a wider range of
formulations of their best-selling drugs, and attempted to switch a
formulation of the product to the OTC sector.
If leading firms switch a post-patent modified version to the OTC
market, then the market will gain in importance (in value terms) relative
to prescribed Pharmaceuticals. For prescription only drugs, marketing is
carried out through two channels: (i) by sales representatives (commonly
known as 'detailers'); and (ii) in medical magazines. In the OTC market,
on the other hand, advertising takes place as in a typical advertising
intensive industry, via the mass media, because the products are sold
directly to the final consumer. As market size increases, this would be
associated with escalating advertising outlays (see Prediction 1), which
work in the same direction as an increase in the cost of innovation, ({>, i.e.,
increasing concentration.
Perhaps more interestingly, as the OTC market expands, this implies
an increase in the effectiveness of advertising that, in turn, implies an
increase in the value of a (Sutton [1996]).^' This can be understood as
follows. In the prescribed Pharmaceuticals sector, leading firms maintain

^^ It used to be the case that for a generic introduction, many of the same trials had to be
replicated to prove safety and efficacy. This slowed down entry. Since this act was passed, a
firm only needs show bioequivalence.
^' In an advertising intensive industry, if the market is defined such that advertising covers
all a firm's products in that market, then this is equivalent to asstuning that the number of
technologies is one (a Type 2 high-a industry).
O Biackwell Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

183

large sales networks. Once a new drug is developed, substantial detailing

occurs. When prescribing doctors have been informed once or even a few
times about the drug's therapeutic effectiveness, any further promotion
may be quite redundant. If a saturation level is reached, then in terms of
the theoretical framework, maintaining a large sales network is analogous
to adding to the exogenous overhead costs paid upon entry (see Sutton
[1991]; also Prediction 3) and there will be no effect on the limiting level of
concentration. This is not the case however within the OTC market. As
long as one deviant firm can convince some proportion of consumers to
buy its product, through escalating its advertising expenditure as
compared to its rivals, then the lower bound to concentration is bounded
away from zero, no matter how large the market becomes (Sutton [1991,
pp. 79-81]). Finally, inasmuch as both advertising and R&D are used as
competitive weapons (assuming they are imperfect substitutes), this will
increase the aggregate overhead costs incurred and so raise equilibrium
concentration (Lyons and Matraves [1996, H3, p. 99]).^^ This yields
Prediction 4.
Prediction 4: Given the level of effectiveness of R&D, a rise in the
effectiveness of advertising implies a rise in a, and hence in the minimal
level of concentration for any given market size.
IV.

EVIDENCE

In this section, evidence is presented in support of Predictions 1-4.

Prediction 1 states that an increase in market size is associated with
escalating advertising and/or R&D expenditure. Prediction 2 states that
concentration must rise if price competition becomes tougher (as long as
advertising/R&D expenditure is constant or increases). Prediction 3 states
that an increase in the cost of innovation, 0, implies an increase in
concentration for any given market size. Prediction 4 states that given the
level of effectiveness of R&D, an increase in the effectiveness of advertising
will increase the value of a, and so the minimal level of concentration.
Consider Prediction 1. Table III shows how expenditure on R&D
and advertising has varied in the UK (and also the US for R&D)
over the past decade. Looking at the evidence on advertising

^^ The example in footnote 2 can be easily generalised to two perceived quality enhancing
expenditures, , and Ej, where

In the limit, as S -> oo, [N + JV"' ] -* [2 + (y^/2(y -I- j8))] < [2 + (y/2)]. Consequently the limiting
level of concentration is higher with two types of endogenous sunk costs.
O Biackwell Publishers Ltd. 1999.

184

CATHERINE MATRAVES
TABLE III
UK AND US FIRM EXPENDITURE ON ADVERTISING AND R&D INDUSTRY (SMILLION)

UK R&D
UK R&D/S
US R&D
US R&D/S
UK Ads
UK Ads/S

1981

1983

1985

1987

1989

1991

1993

594.4
11.3
1866.2
8.3
66.2
1.3

572.7
11.7
2663.1
9.7
65.2
1.3

602.1
11.8
3370.7
10.7
65.5
1.3

1089.9
13.5
4503.2
11.5
111.1
1.7

1513.9
14.7
6019.3
12.3
130.9
1.5

2011.6
15.6
7923.6
13.0
155.2
1.4

2476.0
17.4
10473.0
14.8
166.7
1.4

Source: R&D/S = R&D to sales ratio; Ads/S = advertising to sales ratio; UK: Central Statistical Office
(CSO production data); Register-MEAL (Ads data); CSO Business Monitor MOM (R&D data); US:
OECD (production data); PhRMA (R&D data).

expenditure,^^ observe that the UK advertising to sales ratio has remained

more or less constant, at an average of 1.4%. This implies that as UK
market size has been increasing, advertising expenditure has also been
increasing, which is consistent with Prediction 1. Additionally, substantial
increases in advertising expenditure have been within direct-to-consumer
advertising (Pharmaceutical Online), implying that the effectiveness of
advertising has indeed increased. This supports the first part of Prediction
4}'^ It is when considering firm R&D expenditure that the results are most
striking. The R&D to sales ratio has been increasing over time, implying
that R&D expenditure is increasing faster than market size. The direct
market size effect is then outweighed and we would predict merger and/or
exit, as the least efficient firms would not be able to cover their total costs
in the long run (see Table VII).
Table IV shows the variation in R&D expenditure since 1983 as a
proportion of gross output in the 'Big 4' EU member states, the US and
Japan. In 1983, the lowest R&D to sales ratio was in Japan which was a
protected domestic industry. This can be explained as follows. Given the
smaller protected market initially, Japanese firms had less incentive to

^^ Only UK advertising data are available. However, advertising is typically dependent on

local media, culture and language, and if the theory is correct, the tmderlying competitive
mechanisms would be broadly similar across countries. Thus, if UK advertising expenditure
is increasing over time, we would expect to observe similar trends in other countries; i.e., the
UK is acting as the representative market. Access to both UK and US R&D to sales ratios
allows these to proxy trends in the global market.
^* OTC products account for approximately 20% of the overall market by value, increasing
their market share via the promotion of self-medication for common illnesses such as
headaches, colds and hayfever, and the development of new sectors such as nicotine patches.
1993 global market size was $38.9 billion (US = $11.3; Europe = $9.2; Japan = $8.8). The
1993 advertising to sales ratios were 9.7% for France, 16.3% for Germany, 17.2% for Italy
and 13.4% for the UK (Scrip Yearbook [1995]). In the US, direct advertising to consumers is
now hijgher than promotion to medical journals. The 1995 figures were $356m and $346m
respectively (FT, 24/4/97). By 1996, approximately $600 million was spent, with only 10
firms accounting for more than 90% of the total (Scott-Levin Inc, 10/10/97).
O Biackwell Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

185

TABLE IV
INTERNATIONAL COMPARISON OF FIRM ExPENDrruRE ON

R&D (%)

USA
Japan
Germany
France
Italy
UK

1983

1987

1992

10.6
6.7
8.4
7.1
7.6
11.7

10.6
7.5
9.6
8.4
6.3
13.5

14.3
9.8
9.2
8.7
8.1
16.3

Source: Sharp and Patel,

percentage of gross output.

1996. Total R&D measured as

invest heavily in R&D (assuming high R&D is strongly correlated with

drug discovery), as the market over which such fixed costs could be spread
was relatively small (Sutton [1991]). As globalisation has opened up new
geographical markets to Japanese firms, and also led to more competition
in the Japanese market, this has raised the incentive to escalate R&D in
order to capture a larger global market share.^^ Between 1983 and 1992,
R&D expenditure did increase in Japan. Table IV also shows that the
R&D to sales ratio was initially higher where there was more deregulation
and/or market size was larger, so that expenditure could be spread over
a wider area. The US is both the most deregulated and largest domestic
market, and in absolute terms, R&D expenditure is the highest there.
Overall, the evidence on the increase in both advertising and R&D
expenditure is consistent with Prediction 1.
Now consider Prediction 2 relating to increases in the toughness of
price competition. Table V shows the 1991 price levels in the major EU
member states. Observe first that drugs were relatively cheaper in
France and Spain, and relatively more expensive in the Netherlands and
Denmark. Table V also shows the continuing divergence in
pharmaceutical prices across the EU, due mainly to the differing
institutional social security systems (Panorama [1995]).^^ The variance

"Ballance et al. [Ch. 7, 1992] state that the Ministry of Health and Welfare began to
remove entry restrictions in the 1980s. By 1990, foreign multinationals had a 15% share of the
domestic market. Japanese firms had relied on licensing agreements to sell their products
abroad, but are now beginning to invest abroad. FDI has been growing at 10% annually since
1989 (Scrip Yearbook [1995]). In 1992, Takeda's export sales were 10%, and Sankyo's were
8% of their total sales.
^'There are various market distortions: for example, France and Italy offer higher prices
if firms locate production in the country; even though Spain and the UK do not intervene in
price setting, a maximum level is set on the rate of return. In 1989, the Transparency Directive
was adopted; national authorities must clearly set out their pharmaceutical pricing procedures
(Klepper[1992]).
ID Biackwell Publishers Ud. 1999.

186

CATHERINE MATRAVES
TABLE V

EU PHARMACEUTICAL PRICES 1981-1991

(UNWEIGHTED EU MEAN =

Belgium
Denmark
France
Germany
Italy
NL
Spain
UK

100)

1981

1986

1988

1991

89
*
84
128
78
*
*
121

80

87
126
74
118
87
136
72
110

96
135
63
97
102
132
81
114

69
152
72
140
61
123

Source: Burstall and Senior, 1992.

has, however, decreased over time (see also European Economy [1996,
p. 135]).^^
In the majority of the key markets, there have been attempts to reduce
healthcare costs. For example, prices were cut by 2.5% on reimbursable
drugs in 1993 in the UK, and the increase in fundholding doctors who
control their own budgets put extra pressure on prices. In Germany, the
government introduced a reference pricing system (where drugs are
grouped by therapeutic class) for the first time after reunification, and at
the beginning of 1993, there was a 5% cut on all prices not already
controlled. In Japan, at the beginning of 1993, there was an 8% price cut
on reimbursed drugs. In the US, the rapid growth of HMOs and the
increase in generic substitution has intensified price competition. Although
there continues to be substantial variation in the pricing systems for
reimbursed medicines across the industrialised nations, the evidence
suggests that prescribing doctors must be more accountable, and prescribed drugs more cost-effective. This supports the predicate Prediction 2
where if advertising/R&D expenditure increases and this is combined with
tougher price competition, then concentration must increase.
Predictions 2-4 therefore imply an increase in concentration.
Importantly, the 'correct' geographic market must be considered: if
competition is indeed taking place at the global level, then there should
not exist any systematic pattern at the national level. In other words,
concentration may increase or decrease at the national level, as firms
[1996] provides a cross-country comparison for 1992 weighted average prices
(at the manufacturers' price level); the drugs included are matched on the basis of molecular
content and therapeutic category. Constructing a Laspeyres index based on the price per
kilogram of the active ingredient (weighting by US quantity volumes) shows that prices are
higher in Japan and Switzerland than in the US (28% and 5% resp.), but are considerably
lower in France (43%), Italy (26%) and the UK (32%). Note these types of price comparisons
depend on the sample of drugs selected, the inclusion of generics, and the weighting scheme,
and moreover, are sensitive to the index used.
O Biackwell Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

187

TABLE VI
THE 4-FIRM CONCENTRATION RATIO BY THE VALUE OF PRODUCTION

Germany

CR4

UK

CR4

France

CR4

1987
1991

26
28

1986
1993

34
35

1985
1992

11
11

Italy

CR4

EU

CR4

USA

CR4

1987
1991

17
15

1987
1993

19
16

1979
1986

26
26

Note: The Japanese 1990 5-finn CR is 21.4% (Sharp and Patel, 1996).
Source: German and Italian data are derived from the firm size distributions (resp. by sales and
employment). UK and US (1986) data are derived from their Censuses of Production; French data were
provided by INSEE. Finally, the source for the 1979 US CR4 is IMS America, as reported in Cool, Roller
and Leieux (1998).

locate wherever is efficient. If higher global concentration is, however,

not observed in our data, then given the increase in advertising and R&D
expenditure already shown, combined with tougher price competition, the
evidence would not support Predictions 2-4. Table VI shows the changes
in national (including the EU) concentration since 1987. Observe that
overall, concentration is fairly stable over this time period. At the EU
level, concentration has decreased by 3 percentage points over this seven
year period, which is a relatively large change for a fairly short time
period. Interestingly, it appears that there is no systematic pattern in the
movement in national and EU concentration at the 3-digit level.
Now consider changes in global concentration. Table VII details the
TABLE VII
MERGERS AND ACQUISITIONS (ALL VALtres IN BILLION)

1985
1988
1989
1990
1993
1994

1995

1996
1997

Monsanto (US) and Searle (US).

Eastman Kodak (US) acquired Sterling (US) for S5.1.
SmithKline Beckman (US) and Beecham (UK) merged. Bristol-Myers (US) and
Squibb (US) merged. Dow (Merrell) (US) and Marion (US) merged. American
Home Products (AHP) (US) acquired AH Robins (US) for $3.2.
Rhone-Poulenc (Fra) acquired Rorer (US) for $3.5. Roche (US) bought 60% of
Genentech (US) (biotech firm) for $2.1.
Merck (US) paid $5.9 for Medco (US distributor). Synergen (US) and Amgen
(US) merged ($2.6 billion).
Ciba Geigy (Ch) paid $2.1 for 50% of Chiron (US biotech firm). AHP (US)
acquired American Cyanamid (US) for $9.8. Roche (Ch) acquired Syntex (US)
for $5.1. SmithKline Beecham (UK) paid $2.9 for Sterling Health (US) and resold
part of it to Bayer (Ger) for $1. Eli Lilly (US) paid $9 for PCS (US distributor).
Glaxo (UK) acquired Wellcome (UK) for $ 14. Hoechst (Ger) acquired Marion
Merrell Dow (US) for $7.1. Pharmacia (Swed) and Upjohn (US) merged. RhonePoulenc Rorer (Fr) acquired Fisons (UK) for $1.7 and BASF (Ger) acquired
Boots (UK) for $1.3.
Ciba-Geigy (Ch) and Sandoz (Ch) merged forming Novartis.
Roche (Ch) acquired Boehringer Mannheim (Ger) for $11 billion.

O Biackwell Publishers Ltd. 1999.

188

CATHERINE MATRAVES
TABLE VIII
TOP 20 MARKET SHARES (%) WORLDWIDE: 1983-1995

Company
Glaxo Wellcome (UK)
Merck (US)
Hocchst Marion Roussel'
(Ger)
Bristol-Myers Squibb' (US)
American Home Products
(US)
Pfizer (US)
Johnson & Johnson (US)
Roche (Ch)
SmithKline Beecham' (UK)
Ciba-Geigy (Ch)
Rhone-Poulenc Rorer (Fr)
Bayer (Ger)
Eli Lilly (US)
Sandoz (Ch)
Schering-Plough (US)
Astra (Sweden)
Abbott (US)
Upjohn Pharmacia' (US/Sw)
Sankyo (Jap)
Takeda (Jap)
TOTAL SALES (Sbillion):
TOP 10 SHARE (%)
TOP 20 SHARE (%)

1995
Share

Rank

1992
Share Rank

1988
Share

Rank

1983
Share

Rank

4.5
3.5
3.5

1
2=
2=

3.8
3.6
2.6

1
2
4

2.7
3.2
2.5

2
1
3

1.2
3.1
2.5

17=
1
4

3.1
3.0

4
5

2.8
2.0

3
9=

1.6
2.1

12=
6

1.7
2.6

10=
2=

2.0
1.9
2.1
2.2
2.2
1.8
2.0
2.0
2.1
1.5
1.1
1.8
1.3
1.0
1.5
229.9
25.4
41.6

9=
13
7=
5=

2.9
2.9
2.6
2.5
2.5
2.2
2.1
2.0
1.9
1.9
1.8
1.8
1.7
1.6
1.6
250.0
31.0
49.6

6=
6=
8
9=
9=
11
12
13
14=
14=
16=
16=
18
19=
19=

14=
9=
9=
7=
16=
22=
14=
18
24
16=

1.6
12=
1.5 16=
1.5
16=
1.3 21 =
2.2
5
1.3 21 =
2.3
4
1.7
9=
7=
2.0
1.4
18=
Not Top 25
1.7
9=
1.4
18=
Not Top 25
2.0
7=
156.6
22.4
37.7

2.4
5
1.9
8
1.8
9
2.3
6
2.6
2=
1.1
19=
1.7
10=
2.1
7
1.7
10=
1.4
14
Not Top 25
1.0
21 =
1.3
15=
0.8
24
1.0
21 =
86.7
23.0
36.6

Source: Adapted from Financial Times, 25/3/96; Sharp and Patel, 1996; Ballance et al., 1992; Bogner,
Thomas and McGec, 1996. Total sales are derived from the OECD 'Industrial Structure Statistics'.
Notes: This Table shows the Top 20 for 1995 and how they ranked in 1992, 1988 and 1983.
a Wellcome: 1992=1.2; 1988 = 0.9; 1983 = 1; b Marion Merrell Dow: 1992 = 1.2; c Squibb: 1988 = 1.4;
1983 = 1.1; dBeecham: 1988 = 1.2; 1983 = 1.2; e Pharmacia: 1992 = 1.2.

extent of the recent merger activity in the pharmaceutical industry.

Table VIII shows the changes in global market shares between 1983 and
1995. Since the 1989 merger of SmithKline with Beecham to the Ciba
Geigy-Sandoz merger forming Novartis in 1996, the industry has been
rapidly restructuring itself leading to a consolidation of firms at the top.
Table VII shows that this restructuring has been dominated by global
(inter-regional) activity. There has also been a substantial increase in the
creation of joint ventures: both RJVs, particularly between large
multinational firms and biotech start-ups (Casper and Matraves [1997]),
and marketing joint ventures.
Observe the huge amount of merger activity in 1995. This may be
explained by the fact that R&D works with a lag which has also been
lengthening over time. Given that the average NCE takes between 12 and
15 years to bring to the market, no immediate impact from changes in
R&D expenditure on the level of concentration would be expected.
O Blackwell Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

189

Table IV clearly shows that R&D expenditure escalated from 1987, but
only recently has merger activity been substantial. In the UK, for example,
the number of world-class pharmaceutical firms was reduced from six to
three in 1995. Boots Pharmaceuticals was acquired by BASF, Fisons was
acquired by Rhone-Poulenc Rorer (RPR), and most importantly, Glaxo
and Wellcome merged (forming the world's largest pharmaceutical firm).
Although the firm-level company account evidence shows that it is
currently true that mid-sized firms can compete with the industry's leaders
in a limited line (e.g., Zeneca, UK), some of the recent merger activity,
as shown in Table VII, indicates that a mid-sized firm may be less able to
survive if it wants to be an innovator, (e.g., Fisons/RPR or Upjohn/
Pharmacia).^^ This is the impact of the cost of innovation effect as stated
in Prediction 3.
Table VIII shows the changes in the global market shares of the leading
twenty pharmaceutical firms between 1983 and 1995. Up until 1988,
market shares remained pretty stable although firms were changing rank.
Between 1988 and 1995, the global market shares of the Top 10 firms
increased from 25.4% to 31%, an increase of 5.6 percentage points in
concentration. This is a significant (z = 3.56) increase and also indicates
that given that market size increased over this time period, the largest
firms are growing more quickly than market size. Furthermore, the second
rank of firms (11th to 20th) gained market share, rising from 16.8% in
1992 to 18.6% in 1995." Overall, this significant increase in global
concentration is consistent with Predictions 2-4.
This section has shown that the data support our expectations. It is
interesting to ask, however, what the alternative hypothesis would be. It is
clear that the alternative is not that the pharmaceutical industry is a
Type 1 industry, as in such industries, firms compete in price only. Only
within the generic pharmaceutical sector where firms are commodity
producers, do firms compete in price. As market size increases, we would

^* Alternatively, it may be that due to the wide range of available technologies, the national
and EU data mask changes in therapeutic class concentration. In other words, concentration
could be increasing at the therapeutic class level but could still be decreasing overall if indeed
firms are specialising more. Firm level evidence (derived from company accounts) shows that
of the leading UK firms in 1993, for example, only Glaxo and SKB operated in several
therapeutic classes; Glaxo has a leading position in internal medicine and respiratory; SKB in
anti-infectives and internal medicine. Smaller firms such as Zeneca and Wellcome are far
more specialised; world-wide, Zeneca leads in one technology only, oncology; and Wellcome
(acquired by Glaxo in 1995) in anti-virals.
^' Using a standard t-test, it was found that within the top 10, the difference in mean market
share between 1983 and 1988 or 1992 was insignificant. However, the difference was
significant at the 1% level when comparing 1983 (or 1988) and 1995, and this was also true at
the 10% level when comparing 1992 and 1995. The same significance pattern is observed
within the Top 20.
O Blackwdl Publishers Ltd. 1999.

190

CATHERINE MATRAVES

expect prices to tend to marginal cost as the number of firms become very
large in this segment of the market.'"
In the context of this case, the alternative hypothesis is that the
pharmaceutical industry is a Type 2 high-a rather than low-a industry. The
arguments put forward in Section II concerning the nature of the
technology indicate this is unlikely. If this industry were a Type 2 high-a
industry, the expectation would be that given the observed increase in
advertising and R&D expenditure, combined with tougher price competition, the pharmaceutical industry would become extremely
concentrated, with the new technology of rational drug design displacing
the old. This is not what is being observed. The evidence on concentration
indicates that although the industry is becoming more consolidated, the
two leading firms still only have a global market share of around 4.5%.
This is a high level of fragmentation in an industry characterised by two
types of endogenous sunk costs, and is due to the proliferation mechanism
at work, rather than the escalation mechanism. Only in the OTC market
would we expect the escalation mechanism to dominate and concentration
to be correspondingly higher.
V.

DISCUSSION AND CONCLUDING REMARKS

Various structural changes have occurred in the pharmaceutical industry

in recent years. These include increased international regulatory
harmonisation, governmental attempts to control rising healthcare costs
combined with rapidly ageing populations, and a substantial increase in
the cost of innovation, (j) (mainly through the lengthening development
time). These structural changes have had an important impact on the
competitive process, and given the observed increase in advertising and
R&D expenditure and the toughness of price competition, we predicted
that concentration would increase. This increase would take place not at
the national level, but at the global level. The evidence indeed showed no
systematic pattern in changing market shares at the national and EU level,
but concentration has significantly increased at the global level. Overall,
the basic competitive mechanisms in the pharmaceutical industry are
consistent with the Sutton approach to market structure. The recent
increases in global concentration can be traced primarily to a rise in <p, but
there is an important secondary influence that has worked in the same
direction which is the increasing effectiveness of advertising as the OTC
market expands.
'" Sutton [1991] provides a nice example of the frozen food market, in which a dual market
structure emerged. In the retail sector, firms escalated advertising expenditure as market size
increased, and concentration is high. In the non-retail sector, firms competed on price, and
there are a large number of smaller firms,
O Blackwell Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

191

The fascinating aspect of the competitive process in pharmaceuticals is

why this industry continues to remain relatively fragmented, given that it
is both advertising and R&D intensive. We argued that this is due to the
nature of the technology such that the incentive to proliferate dominates
the escalation mechanism. The pharmaceutical industry generally is a
Type 2 low-a industry. Although the mode of discovery is changing,
through the introduction of biotechnology into the competitive arena, we
argued that the value of a is likely to remain constant over time. This
implies that the pharmaceutical industry is unlikely to become extremely
concentrated. In other words, even if biotechnological processes do win
out, due to the existence of many research trajectories (and associated
therapeutic classes), escalation along one technological trajectory is
unlikely to cause firms on other trajectories to exit. Any changes in the
value of a will be observed through an increase in the effectiveness of
advertising. It is predicted that the escalation mechanism will come into
play in the OTC market, i.e., firms will compete as in a typical Type 2
high-a market, with an escalation of direct-to-consumer advertising in
response to any increases in market size. This may have a substantial
concentrating effect in this sector of the industry.
However, despite the pharmaceutical industry being a Type 2 low-a
industry, there are nevertheless limits to how unconcentrated it can be.
This is due to the existence of a firm size threshold, which has increased
because of the following two factors. First we have seen an increase in the
cost of innovation linked with the skewness of the rate of return on a
new drug. Secondly, a global marketing/distribution network is necessary
in order to exploit a new drug effectively. Both of these factors work
against small firms. Furthermore, these factors are independent of the
nature of the underlying technology in the pharmaceutical industry.
Finally, an interesting avenue for future research is to analyse the role
of firm-specific competencies. These may determine how the different
leading firms react to the underlying Type 2 mechanisms, where the
extemal dynamics of the market and the internal resources of the firm may
combine to determine the R&D expenditure, advertising expenditure and
product line.
ACCEPTED OCTOBER 1998

APPENDIX: THERAPEUTIC CLASSES IN THE PHARMACEUTICAL INDUSTRY

Cardiovascular:
Respiratory system:
Anti-infectives:
O Blackwell Publishers Ltd. 1999.

anti-coagulants, haemophilia, beta-blockers, diuretics, hypertension, cholesterol reducers

anti-histamines, asthma, cough medicines, bronchitis, cystic
fibrosis
antibiotics, antimalarial, anti-virals, vaccines, AIDS

192

CATHERINE MATRAVES

Pain Control:

analgesics, anaesthetics, anti-arthritics, anti-gout, migraine,

bone products (for rheumatism, etc.)
Internal medicine:
antacids, anti-nauseants, contraceptives, enzymes, hormones, laxatives, digestants, anti-ulcerants, immunosuppressants, anti-obesity
Mental Health/CNS: anti-convulsants, sedatives, Parkinson's disease, Alzheimer's,
anti-depressants, multiple sclerosis
Topical:
dermatologicals, haemorrhoids, feminine hygiene preparations, ophthalmic
Cancer Therapy:
cancer therapy (oncology), anti-emesis in cancer treatment
Miscellaneous:
nutrients, vitamins, diabetes, diagnostics

REFERENCES

Ballance, R., Pogany, J. and Forstner, H., 1992, The World's Pharmaceutical
Industries: An International Perspective on Innovation, Competition and Policy,
Edward Elgar, Cheltenham.
Bogner, W., Thomas, H. and McGee, J., 1996, 'A Longitudinal Study of the
Competitive Position and Entry Paths of European Firms in the US
Pharmaceutical Market', Strategic Management Journal, 17, 2, pp. 85-107.
Bond, R. and Lean, D., 1977, 'Sales, Promotion and Product Differentiation in the
Prescription Drug Markets', in Staff Report to the FTC, Washington DC: US
Government.
Bresnahan, T., 1981, 'Departures from Marginal-Cost Pricing in the American
Automobile Industry: Estimates for 1971-1978', Journal of Econometrics, 11, pp.
201-227.
Burstall, M. L. and Senior, I., 1992, Undermining Innovation: Parallel Trade in
Prescription Medicines, IEA Health and Welfare Unit, London.
Casper, S. and Matraves, C, 1997, 'Corporate Governance and Firm Strategy in
the Pharmaceutical Industry', WZB Discussion Paper FSIV 97-20.
Caves, R., Whinston, M. and Hurwitz, M., 1991, 'Patent Expiration, Entry, and
Competition in the US Pharmaceutical Industry', Brookings Papers on Economic
Activity, 1, pp. 1-48.
Cool, K., Roller, L.-H. and Leleux, B., 1998, 'The Relative Impact of Actual and
Potential Rivalry on Firm Profitability in the Pharmaceutical Industry',
forthcoming in Strategic Management Journal.
Danzon, P., 1996, 'The Uses and Abuses of International Price Comparisons', in
Competitive Strategies in the Pharmaceutical Industry, Helms, R. (ed.), American
Enterprise Institute for Public Policy Research, Washington.
Davies, S. W. and Rondi, L., 1996, 'Intra-EU Multinationality of Industries', in
Davies, S. W. and Lyons, B. R. et al.. Industrial Organisation in the European
Union: Structure, Strategy and the Competitive Mechanism, Oxford University
Press, Oxford.
Di Masi, J., Hansen, R., Grabowski, H. and Lasagna, L., 1991, 'Cost of
Innovation in the Pharmaceutical Industry', Journal of Health Economics, 10,
pp. 107-142.
Dranove, D., 1991, 'The Costs of Compliance with the 1962 FDA Amendments',
Journal of Health Economics, 10, pp. 235-238.
Earl-Slater, A., 1993, 'Pharmaceuticals', in European Industries: Structure, Conduct
and Performance, Johnson, P. (ed.), Edward Elgar, Cheltenham.
Blackwell Publishers Ltd. 1999.

MARKET STRUCTURE, R&D AND ADVERTISING

193

European Economy, 1996, Economic Evaluation of the Internal Market, European

Commission.
Grabowski, H. and Vernon, J., 1990, 'A New Look at the Returns and Risks to
Pharmaceutical R&D', Management Science, 36, pp. 804-821.
Grabowski, H. and Vernon, J., 1992, 'Brand Loyalty, Entry, and Price
Competition in Pharmaceuticals after the 1984 Drug Act', Journal of Law and
Economics, 35, pp. 331-350.
Grabowski, H. and Vemon, J., 1994a, 'Innovation and Structural Change in
Pharmaceuticals and Biotechnology', Industrial and Corporate Change, 3, pp.
435-449.
Grabowski, H. and Vernon, J., 1994b, 'Returns to R&D on New Drug
Introductions in the 1980s', Journal of Health Economics, 13, pp. 383-406.
Hansen, R., 1979, 'The pharmaceutical development process: estimates of current
development costs and times and the effects of proposed regulatory changes', in
Chien, R. I. (ed.). Issues in Pharmaceutical Economics, Lexington Books,
Lexington, MA.
Henderson, R. and Cockburn, I., 1996a, 'Scale, Scope and Spillovers: the
Determinants of Research Productivity in Drug Discovery', Rand Journal of
Economics, 27, pp. 32-59.
Henderson, R. and Cockburn, I., 1996b, 'The Determinants of Research
Productivity in Ethical Drug Discovery', in Competitive Strategies in the
Pharmaceutical Industry, Helms, R. (ed.), American Enterprise Institute for
Public Policy Research, Washington.
Hendricks, K., Porter, R. A. and Boudreau, B., 1987, 'Information, Returns and
Bidding Behaviour in OCS Auctions: 1954-1969', Journal of Industrial
Economics, 35, pp. 517-542.
Klepper, G., 1992, 'Pharmaceuticals: Who's Afraid of 1992?', in Winters, Alan,
L. (ed.). Trade Flows and Trade Policy after 1992, Cambridge University Press,
Cambridge.
Lyons, B. R. and Matraves, C , 1996, 'Industrial Concentration', in Davies, S. W.
and Lyons, B. R. et al.. Industrial Organisation in the European Union: Structure,
Strategy and the Competitive Mechanism, Oxford University Press, Oxford.
Lyons, B. R., Matraves, C. and Moffatt, P., 1998, 'Industrial Concentration and
Market Integration in the European Union', Working Paper, WZB.
Panorama of EC Industry, 1995, Commission of the European Communities,
Brussels.
Penan, H., 1996, 'R&D Strategy in a Techno-Economic Network: Alzheimer's
Disease Therapeutic Strategies', Research Policy, 25, pp. 337-358.
PhRMA Industry Profile, 1996 and 1997 issues, Washington DC.
Pisano, G. and Wheelwright, S., 1995, 'The New Logic of High-Tech R&D',
Harvard Business Review, pp. 93-105.
Robinson, W. T. and Chiang, J., 1996, 'Are Sutton's Predictions Robust?
Empirical Insights into Advertising, R&D and Concentration', Journal of
Industrial Economics, pp. 389-408.
Schmalensee, R., 1992, 'Sunk Costs and Market Structure: A Review Article',
Journal of Industrial Economics, pp. 125-134.
Schwartzman, D., 1976, Innovation in the Pharmaceutical Industry, Johns Hopkins
University Press, Baltimore.
Scrip Yearbook, 1995, Scrip Reports, PJB Publications Ltd.
Senker, J., Joly, P.-B. and Reinhard, M., 1996, 'Overseas Biotechnology Research
by Europe's Chemical/Pharmaceuticals Multinationals: Rationale and Implications', Final report for EC Biotech Project, SPRU, University of Sussex.
O Blackwdl Publishers Ltd. 1999.

194

CATHERINE MATRAVES

Shapiro, C, 1989, 'The Theory of Business Strategy', Rand Journal of Economics,

pp. 125-137.
Sharp, M. and Patel, P., 1996, 'Europe's Pharmaceutical Industry: An Innovation
Profile', Draft Report Prepared for DG XU1, European Commission.
Slade, M., 1987, 'Interfirm Rivalry in a Repeated Game: An Empirical Test of
Tacit Collusion', Journal of Industrial Economics, 35, pp. 499-516.
Sutton, J., 1991, Sunk Costs & Market Structure: Price Competition, Advertising,
and the Evolution of Concentration, MIT Press, London.
Sutton, J., 1996, 'Technology and Market Structure', European Economic Review,
pp. 511-530.
Sutton, J., 1998, Technology and Market Structure: Theory and History,
forthcoming, MIT Press.
Symeonidis, G., 1997a, 'Price Competition and Market Structure: the Impact of
Restrictive Practices Legislation on Concentration in the UK', LSE Sticerd
EI/18.
Symeonidis, G., 1997b, 'Cartel Policy, Non-Price Competition and Market
Structure: Theory and Evidence from the UK', LSE Sticerd, El/19.
Temin, P., 1979, 'Technology, Regulation and Market Structure in the Modem
Pharmaceutical Industry', Bell Journal of Economics, 10, pp. 429-446.
Thomas, L. G., 1996, 'Industrial Policy and Intemational Competitiveness in the
Pharmaceutical Industry', in Competitive Strategies in the Pharmaceutical
Industry, Helms, R. (ed.), American Enterprise Institute for Public Policy
Research, Washington.

Blackwell Publishers Ud. 1999.